|
|
 |
| Latent Class Growth Anlaysis Quadrati... |
 
|
| Message/Author |
|
|
| Brian Don posted on Thursday, February 09, 2012 - 11:28 am
|
|
|
I am conducting a latent class growth analysis with four time points and three classes. I originally tested a quadratic term for all classes using the following syntax:
CLASSES = c(3);
Analysis: type = MIXTURE;
STARTS = 100 10;
STITERATIONS = 10;
Model: %OVERALL%
i s q| RST1@0 RST2@3 RST3@6 RST4@11;
i-q@0;
The quadratic term was only significant for one of the classes (e.g. class 1). Thus, I want to retain the quadratic term for Class 1 only and re-run the analysis. However, when I re-run the analysis after fixing the other two classes' quadratic term to zero, the classes seem to arbitrary switch so that Class 1 is no longer Class 1, and the quadratic term is fixed for the wrong two classes. Syntax for the second analysis:
Model: %OVERALL%
i s q| RST1@0 RST2@3 RST3@6 RST4@11;
i-q@0;
%c#1%
[q];
%c#2%
[q@0];
%c#3%
[q@0];
Any help would be appreciated. Thanks! |
|
|
|
|
| The classes can change order. If you don't want this, you need to use starting values and STARTS=0; |
|
|
| Back to top |
|
|
